Performance of Comprehensive Risk Adjustment for the Prediction of In-Hospital Events Using Administrative Healthcare Data: The Queralt Indices

Abstract

Background

Accurate risk adjustment is crucial for healthcare management and benchmarking.

Purpose

We aimed to compare the performance of classic comorbidity functions (Charlson’s and Elixhauser’s), of the All Patients Refined Diagnosis Related Groups (APR-DRG), and of the Queralt Indices, a family of novel, comprehensive comorbidity indices for the prediction of key clinical outcomes in hospitalized patients.

Material and Methods

We conducted an observational, retrospective cohort study using administrative healthcare data from 156,459 hospital discharges in Catalonia (Spain) during 2018. Study outcomes were in-hospital death, long hospital stay, and intensive care unit (ICU) stay. We evaluated the performance of the following indices: Charlson’s and Elixhauser’s functions, Queralt’s Index for secondary hospital discharge diagnoses (Queralt DxS), the overall Queralt’s Index, which includes pre-existing comorbidities, in-hospital complications, and principal discharge diagnosis (Queralt Dx), and the APR-DRG. Discriminative ability was evaluated using the area under the curve (AUC), and measures of goodness of fit were also computed. Subgroup analyses were conducted by principal discharge diagnosis, by age, and type of admission.

Results

Queralt DxS provided relevant risk adjustment information in a larger number of patients compared to Charlson’s and Elixhauser’s functions, and outperformed both for the prediction of the 3 study outcomes. Queralt Dx also outperformed Charlson’s and Elixhauser’s indices, and yielded superior predictive ability and goodness of fit compared to APR-DRG (AUC for in-hospital death 0.95 for Queralt Dx, 0.77–0.93 for all other indices; for ICU stay 0.84 for Queralt Dx, 0.73–0.83 for all other indices). The performance of Queralt DxS was at least as good as that of the APR-DRG in most principal discharge diagnosis subgroups.

Conclusion

Our findings suggest that risk adjustment should go beyond pre-existing comorbidities and include principal discharge diagnoses and in-hospital complications. Validation of comprehensive risk adjustment tools such as the Queralt indices in other settings is needed.

Video abstract

Point your SmartPhone at the code above. If you have a QR code reader the video abstract will appear. Or use:

https://youtu.be/bUzPgB1KMRg

Keywords:

• benchmarking
• case-mix
• comorbidity
• discrimination
• multimorbidity
• Queralt’s indices
• risk

Acknowledgments

The authors would like to thank Drs. Josep Maria Argimon, Vicenç Martinez Ibañez, Ana Ochoa de Echagüen Aguilar, and Jordi Trelis Navarro for their valuable support for the development of the Queralt indices.

Abbreviations

AIC, Akaike’s information criterion; APR-DRG, All Patients Refined Diagnosis Related Groups; AUC, area under the receiver-operator characteristic curve; BIC, Bayesian information criterion; CCS, Clinical Classifications Software; CI, confidence interval; CVD, cardiovascular disease; DRG, Diagnosis Related Groups; ICD-10-CM, International Classification of Diseases, 10th Revision, Clinical Modification; ICS, Catalan Institute of Health (Institut Catala de Salut); ICU, intensive care unit; Queralt DxS, Queralt’s Index for secondary hospital discharge diagnoses, excluding in-hospital, complications; Queralt Dx, Queralt’s Index for secondary hospital discharge diagnoses, including in-hospital complications; ROC, receiver operating characteristic.

Ethics Approval and Informed Consent

The study was approved by the ethics in research committee of the Bellvitge Biomedical Research Institute (IDIBELL). This was a retrospective study, and de-identified data was used. For these reasons, request of written informed consent was not deemed necessary.

Author Contributions

All authors made substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; took part in drafting the article or revising it critically for important intellectual content; gave final approval of the version to be published; and agree to be accountable for all aspects of the work.

Disclosure

David Monterde declares that he is the developer of a software tool that can be used to compute the Queralt Index in clinical, management and research settings. The tool is available online at no cost. Dr Josep Comin-Colet reports grants, personal fees, non-financial support from Vifor Pharma, grants, personal fees from Novartis, grants, personal fees, non-financial support from Orion Pharma, outside the submitted work. The authors declare that they have no other conflicts of interest relevant to the content of this manuscript.